Symbolic Graphic Emotion Design in Gift Packaging Based on Multimodal Emotion Fusion

Abstract

This paper firstly discusses the value significance of emotional packaging design, outlines its design principles and methods, and explains the thinking structure of emotional design. Secondly, the feature detection of symbolic graphics is realized using a mixture of regression and segmentation, and a multilevel cross-modal perceptual emotion recognition model is constructed based on the collaborative attention mechanism of the cross-modal Transformer and recurrent neural network. Finally, a dataset test and empirical analysis were conducted to verify the model’s effectiveness in this paper. The results show that the multimodal fusion model has better recognition performance than the unimodal and bimodal models, with F1 values of 0.914, 0.923, and 0.902 for neutral, positive, and negative emotions, respectively. Symbolic graphics in the emotional design of gift packages are mainly expressed in the form of intuition, vividness, and validity to emphasize the diversity of symbolic graphics in the emotional design of gift packages. This shows that multimodal emotional fusion can be realized to analyze the emotional design of symbolic graphics of gift packaging and promote the emotional innovation design of symbolic graphics of gift packaging.

Keywords

• multimodality
• symbolic graphics
• collaborative attention mechanisms
• recurrent neural networks
• affective design
• 65d17